Scale-up of AAV production in iCELLis™ fixed-bed bioreactors

The iCELLis™ bioreactor system is a fully automated, single-use, fixed-bed bioreactor engineered to support efficient, adherent cell culture processes and high-yield viral vector production. Its fixed-bed architecture offers an expansive surface area for cell growth within a compact footprint, making it an excellent choice for scaling up adherent cell processes and manufacturing viral vectors used in gene therapies and vaccines.

The iCELLis™ Nano and 500+ bioreactors have been commercially available for over a decade. The iCELLis™ 500+ bioreactor is currently being successfully utilized in the production of six approved gene therapies, including Zolgensma, an AAV9-based therapy for spinal muscular atrophy (SMA).

The iCELLis™ 50 bioreactor completes the iCELLis™ bioreactor family (Fig 1) by providing a geometrically consistent scale-down model of the iCELLis™ 500+ bioreactor, with surface areas ranging from 6 to 50 mm². This makes it well-suited for intermediate-scale applications, including small phase 1 and phase 2 clinical batches.

This application note aims to show the consistent and reliable performance of an rAAV production process across the iCELLis™ Nano, iCELLis™ 50, and iCELLis™ 500+ bioreactor scales.

Fig 1. The iCELLis™ family of fixed-bed bioreactors: iCELLis™ Nano, 0.53 to 4 m² (left), iCELLis™ 50, 6 to 50 m² (middle), iCELLis™ 500+, 66 to 500 m² (right).

We successfully completed rAAV5 production runs across all three scales of the iCELLis™ bioreactor family, utilizing vessels with 10 cm bed heights and low compaction. These configurations provided available surface areas of; 2.65 m² for the iCELLis™ Nano bioreactor, 33 m² for the iCELLis™ 50 bioreactor, and 333 m² for the iCELLis™ 500+ bioreactor. More information about the different fixed-bed sizes can be found in the Appendix. Across all scales, the iCELLis™ bioreactors exhibited robust performance, with consistent cell growth and metabolic activity during rAAV5 production. The process yielded average titers of 1.2 × 10¹⁰, 9.2 × 10⁹, and 9.7 × 10⁹ viral genomes (vg) per cm² of cell growth surface area in the Nano, 50, and 500+ bioreactors, respectively, demonstrating reliable productivity throughout the bioreactor range.

Scaling across the iCELLis™ bioreactor platform requires the consideration of multiple variables. The vertical velocity of the media flowing upwards through the fixed-bed (linear speed) and the media volume-to-cell growth surface area (V/A) ratio can be critical for maintaining a homogenous environment conducive to consistent cell growth and productivity. The height of the media flowing downward from the top of the fixed-bed (falling film) can affect the efficiency of the low-shear gas exchange. Parameter selection should align with the priority of the targeted process step and is considered within the scope of this study.

Experimental work was performed from May 2025 to October 2025 at Cytiva, Westborough.

Table 1. Equipment used in this study

Table 2. Plasmids and reagents used in this study

Table 3. Consumables used in this study

Production cohort order

Four distinct production cohorts were conducted to collect data across the full range of iCELLis™ bioreactor scales, completing 10 production batches in total. For batches within each cohort, all cell culture procedures were performed simultaneously, utilizing shared material pools to minimize potential sources of external variability. These pooled materials included cells for bioreactor inoculation, prepared media and solutions, and transfection complexes. The sequence of bioreactor data generation is detailed in Table 4.

Table 4. Quantity of iCELLis™ bioreactor batches completed in each production cohort

Media composition and preparation

Two media formulations were utilized throughout testing:

1. Complete media was composed of high-glucose HyClone™ Dulbecco’s modified eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2% GlutaMAX, and 1% MEM non-essential amino acids (MEM NEAA).
2. Serum-free (SF) media was composed of high-glucose DMEM supplemented with 2% GlutaMAX, and 1% MEM NEAA.

Complete media was used for seed train expansion and the cell growth phase in the bioreactor, while SF media was used for the media exchange prior to transfection and the subsequent production phase in the bioreactor.

Each production cohort had its own dedicated media batch, with all vessels within a given cohort using media from that same batch. All media batches were prepared using sterile manipulations of liquid components.

For production cohort 4, the media batches underwent an additional filtration step using a 0.2 µm filter prior to use, addressing the increased risk associated with the larger required volumes.

Seed train expansion

HEK293-H cells were initially thawed from a research-use-only cell bank (RCB) into a single-layer CellSTACK culture chamber containing complete medium. The cells were incubated at 37⁰C in a humidified atmosphere containing 5% CO₂, and expanded in exponential phase through successive passaging every 3 to 4 d, maintaining a total passage count between four and six passages before bioreactor inoculation. For each production cohort, multiple 10-layer chambers were pooled and distributed across bioreactor scales to provide sufficient cell quantities for bioreactor seeding.

Selection of linear speeds and falling film heights

Linear speed refers to the vertical velocity of liquid as it flows upward through the fixed-bed. The selection of the linear speeds of the media being circulated through the fixed-bed during testing was adopted from current Cytiva guidance recommendations for both the iCELLis™ Nano and 500+ bioreactors. Since linear speed characterization studies for the iCELLis™ 50 bioreactor were not finalized prior to the start of scalability testing, the agitation setpoints and vessel volumes used throughout testing were based on preliminary information.

The linear speeds listed in Table 5 to 8 and Table 10 for the iCELLis™ 50 bioreactor were therefore retrospectively calculated from finalized linear speed characterization data. The overall intention for the iCELLis™ 50 bioreactor agitation setpoints was to achieve linear speeds comparable to those of the iCELLis™ 500+ bioreactor.

Likewise, the falling film heights applied during testing of the iCELLis™ 500+ bioreactor were based on prior operational experience and current Cytiva guidance. The falling film height refers to the difference in level between the top of the fixed-bed and the liquid level in the outer chamber of the vessel. For the scalable process in the iCELLis™ Nano and 50 bioreactors, the approach was to either replicate the falling film height of the 500+ bioreactor or align more closely with the media volume-to-cell growth surface area (V/A) ratio, depending on the priority of a targeted process step.

Setpoints for dissolved oxygen control

The location of the dissolved oxygen (DO) sensor varies between the iCELLis™ bioreactors. The iCELLis™ Nano bioreactor has a single sensor located in the media reservoir outside of the fixed-bed. This sensor measures the DO after the media has been oxygenated by the falling film and before contact with the oxygen-consuming cells in the fixed-bed (hereafter referred to as “before the fixed-bed”).

The iCELLis™ 500+ bioreactor has two different DO sensor locations; one sensor is located before the fixed-bed, while the other sensor is located right above the fixed-bed, where it measures DO after contact with oxygen-consuming cells (hereafter referred to as “after the fixed-bed”). In the iCELLis™ 500+ bioreactor, we recommend control of the DO with the sensor located after the fixed-bed. The iCELLis™ 50 bioreactor has duplicate DO sensors that are both located after the fixed-bed.

Oxygen consumption by the cells in the fixed-bed can create a significant difference between the DO level of the media before and after the fixed-bed. Previous studies with the iCELLis™ Nano have shown a 30% reduction in DO across the fixed-bed for the process used during this study (unpublished data). To mitigate this difference, the DO setpoint of the iCELLis™ Nano bioreactor was adjusted to 80% while the 50 and 500+ bioreactors were controlled at a lower setpoint of 50%. All systems utilized single-sided control with addition of oxygen to stay above the setpoint when required.

Process description: Bioreactor media conditioning

Complete medium was added to the bioreactor 1 to 2 d prior to inoculation for vessel conditioning at 10% CO₂. The control parameters used during media conditioning at each bioreactor scale are outlined in Table 5.

A falling film height of 0 cm was used during media conditioning and inoculation for both the iCELLis™ 50 and 500+ bioreactors to minimize expected foam generation, which could entrap seeded cells. In the iCELLis™ Nano vessel, where foam formation is not anticipated, inoculation was performed at the minimum working volume to better replicate the V/A ratio, and thereby the inoculum's volumetric concentration, of the iCELLis™ 500+ bioreactor. This approach resulted in a falling film height of 7 cm for the iCELLis™ Nano bioreactor (Table 5). Given the minimal foam formation observed in the iCELLis™ 50 bioreactor, it is recommended that future production runs also use a minimum working volume during inoculation, to better align with the V/A ratio of the iCELLis™ 500+ bioreactor. This approach would yield a falling film height of 10 cm for the iCELLis™ 50 bioreactor.

Since linear speed characterization studies for the iCELLis™ 50 bioreactor were not finalized prior to the start of scalability testing, the agitation rate of 222 RPM used with the iCELLis™ 50 bioreactor at the 0 cm fixed-bed height resulted in a retrospectively calculated linear speed of 1.4 cm/s, slightly higher than the 1.3 cm/s recommended to achieve linear speeds comparable to the iCELLis™ 500+ bioreactor.

Table 5. Bioreactor setpoints and parameters used during media conditioning

^*For future processes, using the minimum working volume is recommended during conditioning with the iCELLis™ Nano and 50 bioreactors for scaling to the iCELLis™ 500+ bioreactor.
^†For future processes, a linear speed of 1.3 cm/s is recommended during conditioning of the iCELLis™ 50 bioreactor.

Process description: Bioreactor inoculation

For each production cohort, bioreactors were inoculated using the same cell pool. The inoculum was introduced into the bioreactors via the inoculation port by gently pressurizing the prefilled inoculation manifold. No additional flush of the manifold was performed. Following inoculation, cells were given a minimum of two hours to adhere to the fixed-bed prior to initiating media recirculation with complete media. The control parameters used during inoculation and cell attachment at each bioreactor scale are outlined in Table 6.

Table 6. Bioreactor setpoints and parameters used during inoculation

^*For future processes, using the minimum working volume is recommended during inoculation of the iCELLis™ Nano and 50 bioreactors for scaling to the iCELLis™ 500+ bioreactor.
^†For future processes, a linear speed of 1.3 cm/s is recommended during inoculation of the iCELLis™ 50 bioreactor.

Process description: Bioreactor growth and production

After inoculation and cell attachment, the cultures were provided a 5 d growth period before media exchange to serum-free (SF) medium and transfection were performed, as outlined in the next section. Production then continued in SF medium for an additional 5 d prior to initiating harvest. The control parameters used for each bioreactor scale during growth and production are outlined in Table 7.

Because the total media volume required for cell growth and production (0.17 mL of media/cm² of cell growth surface area) exceeded the vessel's internal capacity, media was continuously recirculated from an external biocontainer throughout the growth and production phases at a rate equivalent to 2.5 times the total vessel volume per day.

Table 7. Bioreactor setpoints and parameters used during growth and production

Process description: Bioreactor triple transfection

Transfection was performed after 5 d of cell growth in complete media. To prepare for transfection, media recirculation was halted, and the complete media in each vessel was removed via draining and then replaced with a single vessel volume of preheated serum-free (SF) media. The spent complete media was discarded and not used in any subsequent processing. The control parameters used for each bioreactor scale during transfection are outlined in Table 8.

Transfection was conducted using a triple transfection protocol. The transfection complex consisted of 0.2 µg/cm² total plasmid DNA combined with PEI MAX at a 1:2 ratio, prepared in SF media. A transfection complex pool containing 40 mg/L of DNA was prepared collectively for all batches in the production cohort. After a 15 min incubation, the pool was divided and immediately introduced into each bioreactor using manifold pressurization, with volumes adjusted according to the cell growth surface area to deliver 0.2 µg of DNA per cm². Cultures were given a minimum of 2 h for transfection post complex addition prior to initiating recirculation of SF media.

A falling film height of 0 cm was also used for the iCELLis™ 500+ bioreactor to minimize expected foam generation during transfection and lysis.

The iCELLis™ Nano and 50 bioreactors both minimized their working volumes, with resulting falling film heights of 7 and 10 cm, respectively, to more closely match the V/A ratio, and consequently the volumetric concentrations, of the transfection complex and lysis environment of the iCELLis™ 500+ bioreactor. The agitation rate of 222 rpm used with the iCELLis™ 50 bioreactor at the 10 cm fixed-bed height only resulted in a retrospectively calculated linear speed of 0.7 cm/s. A higher linear speed of 1.3 cm/s is recommended during transfection and lysis in future production runs to achieve linear speeds comparable to those of the iCELLis™ 500+ bioreactor.

Table 8. Bioreactor setpoints and parameters used during transfection

^*For future processes, a linear speed of 1.3 cm/s is recommended during transfection of the iCELLis™ 50 bioreactor

Process description: Bioreactor lysis and harvest

Harvest was initiated 5 d post-transfection. Spent media supernatant was drained from each bioreactor vessel and combined with the spent serum-free media in its external recirculation biocontainer. Lysis buffer supplemented with 25 U/mL Denarase was then added into each vessel and incubated at room temperature overnight. The lysis buffer composition is outlined in Table 9.

Table 9. Lysis buffer composition

The next day, an additional 5 M sodium chloride (NaCl) solution was added to the vessel to a final concentration of 500 mM in the lysate. Following a 30 min incubation period at room temperature, the lysate was drained from each vessel. The vessel was then washed with 1× HyClone™ DPBS for 30 min at room temperature to remove any remaining virus. The final harvest pool consisted of spent media supernatant, lysate, and PBS wash fractions. Both individual harvest fraction samples and final pool samples were stored at -80°C prior to analytical evaluation.

Table 10 outlines the bioreactor setpoints and parameters used during lysis.

Table 10. Bioreactor setpoints and parameters used during lysis

^*For future processes, a linear speed of 1.3 cm/s is recommended during lysis of the iCELLis™ 50 bioreactor.
^†When draining, a constant air flow rate is recommended to avoid creating negative pressure in the vessel

Analytical methods

Metabolite and offline pH and gas measurements:

• Metabolite concentrations (glucose, lactate, glutamine, glutamate, and ammonium), along with pH and gas (pO₂ and pCO₂) levels, were measured offline using the BioProfile Flex 2 cell culture analyzer.
• Single sample volumes were removed and analyzed daily from each bioreactor.

Nuclei counting:

• Carriers were aseptically removed from each iCELLis™ Nano bioreactor prior to transfection and measured for total nuclei counts using the NucleoCounter NC-200 cell counter. The cells on the carriers were lysed and subsequently analyzed using the count of aggregated cells A100 and B Assay protocol on the NC-200 software.
• Three carriers were individually analyzed and then averaged to obtain the final daily cell count.

Viral genome titer measurement:

• We measured viral genome titers using droplet digital PCR (ddPCR) with the Bio-Rad QX200 AutoDG droplet digital PCR system. Samples were pretreated with DNase I followed by dilution prior to droplet generation. We quantitated using a custom primer and probe set targeting our gene of interest, green fluorescent protein (GFP).
• The data shown in the results section represent the mean and standard deviation of single sample aliquots across multiple bioreactors and does not denote assay-specific variability.

Capsid titer measurement:

• Intact capsid titers were determined using the PRAAV5XP quantitative ELISA kit (PROGEN), following the manufacturer’s instructions. This sandwich ELISA method enables accurate quantification of rAAV5 particles.
• The percentage of full capsids was determined via a calculation between the viral genome and capsid titer values.
• The data shown in the results section represent the mean and standard deviation of single sample aliquots across multiple bioreactors and does not denote assay-specific variability.

We were unable to perform further testing for AAV impurities such as residual plasmid DNA.

Harvest titer

Bioreactors were harvested after a 5 d production phase. The spent media supernatant, cell lysate, and PBS wash fractions were collected and combined to obtain the final harvest pool material. Harvest fractions were analyzed for genomic titers only, while harvest final pools were analyzed for genomic and capsid titer via ddPCR and ELISA, respectively.

Fig 2. Comparison of rAAV5 titers from harvest fraction samples (individual supernatant, lysate, and wash fractions) in iCELLis™ Nano (n = 6), iCELLis™ 50 (n = 3), and iCELLis™ 500+ (n = 1) bioreactors. Data shows mean values ± standard deviation

Harvest fraction viral genome titers are shown in Figure 2. All three systems show their highest titers in the lysate fraction (approximately 60% of the total titer), with minimal recovery in the PBS wash fraction (approximately 5% of the total titer). Figure 3 shows the total viral particles (vp) and viral genomes (vg) in the pooled harvest for each bioreactor scale. While overall fraction percentages showed some variation between bioreactor scales, the genomic titer averages from final harvest pools remained consistent across all scales:

• 1.2 ± 0.1 × 10¹⁰ vg/cm² (iCELLis™ Nano bioreactor)
• 9.2 ± 0.9 × 10⁹ vg/cm² (iCELLis™ 50 bioreactor)
• 9.7 × 10⁹ vg/cm² (iCELLis™ 500+ bioreactor)

Fig 3. Comparison of rAAV5 titer and packaging yields from harvest pool samples (combined supernatant, lysate, and wash fractions) in iCELLis™ Nano (n = 6), iCELLis™ 50 (n = 3), and iCELLis™ 500+ (n = 1) bioreactors. Data shows mean values ± standard deviation

A lower viral particle count, accompanied by a higher proportion of full capsids, was observed in the iCELLis™ 500+ bioreactor. However, the source of this variability cannot be determined based on data from a single production batch.

Production run trends

The rAAV5 adherent production process consisted of an initial growth phase, a media exchange and transfection on day 5, and five additional days dedicated to production prior to harvest. Bioreactor supernatant samples were removed daily to obtain offline data on gas exchange and metabolic trends over the course of each cell culture run. All figures in this section show culture trend results across the iCELLis™ bioreactor scales. Trend data for each bioreactor run is displayed separately to show the variability experienced across scales.

Production run gas exchange trends

Bioreactor culture gas exchange trends for pH, DO, and pCO₂ are shown in Figures 4, 5, and 6, respectively. The pH and pCO₂ trends were similar across cultures, with the singular iCELLis™ 500+ bioreactor data set integrating into the variability seen from the other scales. The deviations in pH value around days 6 and 7 for production run 3 in the iCELLis™ 50 bioreactor were due to an overnight recirculation malfunction (Fig 4). pCO₂ data was not available for this data timepoint (Fig 6).

Fig 4. Continuous online pH trends for culture growth and production phases in iCELLis™ Nano (n = 5, gray), iCELLis™ 50 (n = 3, green), and iCELLis™ 500+ (n = 1, blue) bioreactors. The trend from one iCELLis™ Nano batch in cohort one is excluded due to probe-specific deviations that do not accurately reflect the process data. On day 5, a media exchange to serum-free media was completed, followed by transfection. Significant spikes in pH were observed on day 5, attributed to sensor dryness at the time of the media exchange. The drop in pH around day 6 and 7 for one iCELLis™ 50 bioreactor was due to overnight recirculation malfunctions. Bioreactor lysis started on day 10.

The iCELLis™ Nano bioreactor was controlled to a DO setpoint of 80%, while the 50 and 500+ bioreactors were controlled at a lower setpoint of 50% due to the positioning of their DO sensors relative to the fixed-bed structure.

Online DO values for all bioreactors are shown in Figure 5. At the start of the culture, the DO in the 500+ bioreactor was controlled by the sensor before the fixed-bed, an error that was corrected on day 4. Subsequently, measurements taken after the fixed-bed in the iCELLis™ 500+ bioreactor closely resemble those from the iCELLis™ 50 bioreactor, while measurements taken before the fixed-bed align with those from the iCELLis™ Nano bioreactor. This pattern highlights the impact of cellular oxygen consumption within the fixed-bed.

Fig 5. Continuous online dissolved oxygen (DO) trends for culture growth and production phases in iCELLis™ Nano (n = 6, gray), iCELLis™ 50 (n = 3, green), and iCELLis™ 500+ (n = 1, blue) bioreactors. The iCELLis™ 500+ bioreactor displays two distinct trends depending on the placement of the DO sensor, either before the fixed-bed (dark blue) or after the fixed-bed (light blue). Day 0–4: DO control with before fixed-bed sensor, Day 4–10: DO control with after fixed-bed sensor. On day 5, a media exchange to serum-free media was completed, followed by transfection. Significant spikes in DO were observed on day 5, attributed to sensor dryness at the time of the media exchange. Bioreactor lysis started on day 10.

Fig 6. Offline pCO2 trends for culture growth and production phases in iCELLis™ Nano (n = 6, gray), iCELLis™ 50 (n = 3, green), and iCELLis™ 500+ (n = 1, blue) bioreactors. Media exchange to serum-free media along with transfection was completed on day 5.

Production run biomass trends

Bioreactor biomass capacitance trends are shown in Figure 7. Biomass sensors provided real-time cell density monitoring for each bioreactor run, as cell counting of carriers is not available in the iCELLis™ 50 and 500+ bioreactors.

The data was normalized to a percentage of the maximum capacitance value of each batch. Spikes observed in the data were due to intentional agitation stoppages to facilitate manipulations such as Nano fixed-bed carrier sampling, media exchanges, and transfection on day 5. The trends all show a consistent increase in capacitance during cell growth with a flattening in values following transfection.

Fig 7. Continuous online biomass capacitance trends normalized to percent of maximum value per batch for culture growth and production phases in iCELLis™ Nano (n = 5, gray), iCELLis™ 50 (n = 3, green), and iCELLis™ 500+ (n = 1, blue) bioreactors. Spikes observed are an artifact of intentional agitation disruptions. On day 5, a media exchange to serum-free media was completed, followed by transfection. Bioreactor lysis started on day 10.

Production run metabolite trends

Bioreactor culture metabolite trends for glucose, lactate, glutamine, glutamate, and ammonia are shown in Figures 8 through 12, respectively. The trends observed were consistent across all metabolites and bioreactor sizes, demonstrating effective scalability throughout the different scales. Low levels of lactate and ammonium throughout the cultures also suggest favorable conditions for cell growth and productivity. The deviation in metabolite values shown around day 7 for one iCELLis™ 50 bioreactor batch was due to an overnight recirculation malfunction.

Additionally, glucose trends served as an indirect indicator of cell culture growth for each bioreactor run, as direct cell counting of carriers is not available in the iCELLis™ 50 and 500+ bioreactors. Nano bioreactor fixed-bed carriers were sampled prior to transfection; with day 5 cell counts ranging between 117 000 and 129 000 cells per cm² across production runs. The consistent glucose trends during cell growth (Fig 8), combined with the precise cell counts observed in the Nano bioreactor, suggest that cell densities were comparable across all systems at the time of transfection.

Fig 8. Glucose trends for culture growth and production phases in iCELLis™ Nano (n = 6, gray), iCELLis™ 50 (n = 3, green), and iCELLis™ 500+ (n = 1, blue) bioreactors Media exchange to serum-free media along with transfection was completed on day 5. The drop in glucose around day 7 for one iCELLis™ 50 bioreactor was due to an overnight recirculation malfunction.

Fig 9. Lactate trends for culture growth and production phases in iCELLis™ Nano (n = 6, gray), iCELLis™ 50 (n = 3, green), and iCELLis™ 500+ (n = 1, blue) bioreactors. Media exchange to serum-free media along with transfection was completed on day 5. The spike in lactate around day 7 for one iCELLis™ 50 bioreactor was due to an overnight recirculation malfunction.

Fig 10. Glutamine trends for culture growth and production phases in iCELLis™ Nano (n = 6, gray), iCELLis™ 50 (n = 3, green), and iCELLis™ 500+ (n = 1, blue) bioreactors. Media exchange to serum-free media along with transfection was completed on day 5.

Fig 11. Glutamate trends for culture growth and production phases in iCELLis™ Nano (n = 6, gray), iCELLis™ 50 (n = 3, green), and iCELLis™ 500+ (n = 1, blue) bioreactors. Media exchange to serum-free media along with transfection was completed on day 5. The drop in glutamate around day 7 for one iCELLis™ 50 bioreactor was due to an overnight recirculation malfunction.

Fig 12. Ammonium trends for culture growth and production phases in iCELLis™ Nano (n = 6, gray), iCELLis™ 50 (n = 3, green), and iCELLis™ 500+ (n = 1, blue) bioreactors. Media exchange to serum-free media along with transfection was completed on day 5. The spike in ammonium around day 7 for one iCELLis™ 50 bioreactor was due to an overnight recirculation malfunction.

The iCELLis™ 50 bioreactor is the newest addition to the iCELLis™ bioreactor family. This intermediate-scale system, offering cell growth surface areas ranging from 6 to 50 m2, is a geometrically consistent scale-down model of the iCELLis™ 500+ bioreactor. The iCELLis™ bioreactor family now provides a comprehensive selection of surface area configurations from the laboratory to commercial manufacturing, spanning from 0.5 to 500 m².

To evaluate scalability across the iCELLis™ bioreactor platform, rAAV5 production was performed using vessels with cell growth surface areas of 2.65 m² in the Nano bioreactor, 33 m² in the 50 bioreactor, and 333 m² in the 500+ bioreactor. Vessels with 10 cm fixed-bed heights and low compaction configurations were chosen to allow for consistent process assessment across the range of production scales.

We employed two scaling approaches for these iCELLis™ 50 bioreactor batches:

1. Inoculation, cell growth, and production—replicate the falling film height of the iCELLis™ 500 bioreactor
2. Transfection and lysis—align more closely with the media volume-to-cell growth surface area ratio of the iCELLis™ 500 bioreactor

These strategies worked well for this equipment. However, given that minimal foaming was observed in the iCELLis™ 50 bioreactor, the recommendation would be to align with the media volume-to-cell growth surface area ratio, and not falling film height, during inoculation, transfection and lysis. Linear speed targets for the iCELLis™ 50 and 500 bioreactors should also be kept identical between both systems.

Across all scales, the iCELLis™ bioreactors demonstrated consistent performance throughout cell growth and rAAV5 production. Metabolite trends, including glucose, lactate, glutamine, glutamate, and ammonia, remained stable across the Nano, 50, and 500+ scales. Similar pH and pCO₂ patterns were also observed across bioreactor scales, while DO profiles varied due to differences in sensor placement and control strategy. Biomass capacitance profiles, when normalized as a percentage of their maximum value, showed consistent cell density increases during cell growth with subsequent flattening after transfection. Additionally, the iCELLis™ Nano nuclei counts at the time of transfection, which ranged between 117 000 and 129 000 cells per cm² across six production batches, coupled with capacitance and glucose levels at the time of transfection, further suggest that similar cell densities were maintained across all systems at the time of transfection.

Harvest fractions from each production run were analyzed for genomic titers only, while harvest final pools were analyzed for genomic and capsid titer via ddPCR and ELISA, respectively. Some variability in fraction yields, viral particle counts, and full capsid percentages was noted, though the underlying cause remains unclear and may not be industrially relevant. Perceived differences may stem from the low sample size (n) and could potentially be reduced by including additional biological replicates in future experiments. Genomic titer averages from final harvest pools remained consistent across all scales, indicating consistent and dependable productivity across the entire bioreactor scale range.

• 1.2 ± 0.1 × 10¹⁰ vg/cm² (iCELLis™ Nano bioreactor)
• 9.2 ± 0.9 × 10⁹ vg/cm² (iCELLis™ 50 bioreactor)
• 9.7 × 10⁹ vg/cm² (iCELLis™ 500+ bioreactor)

Overall, the process shows consistent output across batches and scales, reinforcing confidence in its scalability and operational robustness.

The iCELLis™ bioreactor family now provides a comprehensive selection of surface area configurations, spanning from 0.5 to 500 m², as detailed in Table 11.

Each iCELLis™ bioreactor vessel is available in six cell growth surface area configurations, achieved by combining three fixed bed heights with both low and high bed density (compaction) levels. Compared to traditional 2D flatware systems, the iCELLis™ bioreactor enables much larger batch volumes while minimizing operational costs and facility space requirements. Additionally, the system complies with 21 CFR Part 11 regulations, ensuring seamless integration into GMP-compliant environments.

Table 11. Fixed-bed sizes available for the iCELLis™ bioreactor family. Fixed-bed sizes are achieved by varying bed height and carrier compaction levels